[Top] [All Lists]

Re: [ontolog-forum] Why a data model does not an ontology make

To: "[ontolog-forum]" <ontolog-forum@xxxxxxxxxxxxxxxx>
From: William Frank <williamf.frank@xxxxxxxxx>
Date: Sun, 31 Mar 2013 10:56:05 -0400
Message-id: <CALuUwtBT=cofLn153-Vd3VQiJuWHe_iYo=MgAjKoxzw9MYXRRg@xxxxxxxxxxxxxx>

On Sun, Mar 31, 2013 at 9:55 AM, John F Sowa <sowa@xxxxxxxxxxx> wrote:
Michel, Leo, and Michael,

E-R-A + cardinality is a requirement that must be specified in any
conceptual schema (or ontology),
Indeed some *means* of identifying what in E/R (and UML, and OWL ...) modelling were distinguished as 'Entities', 'Relationships', 'Attributes', and 'cardinality' (or multiplicity) is a requirement for an ontology language. 

But, I am without doubt  that E-R-A need not be distinguished in the language as the three fundamental and entirely different grammatical and semantic categories: that they are best  ALL reiffied as things, (entities, or objects, or whatever, that can be the values of variables).  The difference between an 'E', like George, a domain-meaningful  'R' (such as is married to), and an 'A' (such as sex) is a feature of many, but not all, natural languages.  (so too the oft-here-debated difference between individuals and categories, whatever we wish to call those).

 Accepting this ontological presupposition of these languages and the legacy of E/R modelling is *inimical* to

     diversity, heterogeneity, and interoperability.

     technology-solution independent models,

     technology solutions not based on the E/R paradigm, such as columnar databases, functional programming ....

Moreover, while a means of specifying multiplicities is important, multiplicities are seldom stable over long periods of time or even across micro-domains. Often, they are simply conveniences of the moment.
(how many users does a  tablet have?) I have found that nothing slows up modelling like arguing over multiplicities, and that baking them into a specification is a recipe for brittleness.   They are better treated as independent constraints, enabling underspecifying the ontology, and following the Kanban principle of deferring commitments to details.

For example, if 'marriage' is defined in an ontology as a thing of the type that has roles attached, and to which the roles of spouse and officiator attach, with any multiplicities or constraints supplied separately, we would have a concept useful across all cultures, while the constraints differ.   And, since people *do* seem to be able to identify 'marriage' as a common concept, even as those constraints vary, so they are interoperating while permitting diversity, in a manner that we would want a semantic web to be able to support. 

So, I would say that a data model does not an ontology make for an even deeper reason, that data modelling is fixed on modelling in a way that is geared toward particular technology solutions: relational databases and O-O programming, and, if fully specified, is fixed on specifying the ephemeral needs of an application.


Message Archives: http://ontolog.cim3.net/forum/ontolog-forum/  
Config Subscr: http://ontolog.cim3.net/mailman/listinfo/ontolog-forum/  
Unsubscribe: mailto:ontolog-forum-leave@xxxxxxxxxxxxxxxx
Shared Files: http://ontolog.cim3.net/file/
Community Wiki: http://ontolog.cim3.net/wiki/ 
To join: http://ontolog.cim3.net/cgi-bin/wiki.pl?WikiHomePage#nid1J    (01)

<Prev in Thread] Current Thread [Next in Thread>